A screw axis identification method for serial robot calibration based on the POE model

The purpose of this paper is to propose a screw axis identification (SAI) method based on the product of exponentials (POE) model, which is concerned with calibrating a serial robot with m joints equipped with a stereo‐camera vision system.

Different from conventional approaches, like the circle point analysis (CPA) or the system theoretic method which must collect a great deal of data, the identification of the joint parameters for the proposed method only needs to measure m+1 times for n (n≥3) target points mounted on the manipulator end‐effector.

In this approach, the joint parameter, called a screw or twist, together with the actual value of joint angle can be obtained by linearly solving a closed‐form expression. Further, this method avoids calibrating the hand‐eye relationship and the exterior parameter of the robot.

Finally, the stability and accuracy of the SAI method are evaluated by simulation experiments, and it is also verified well in practical experiments.